Optically monitoring voltage in neurons by photo-
induced electron transfer through molecular wires
Evan W. Millera, John Y. Lina, E. Paxon Fradyb, Paul A. Steinbacha,c, William B. Kristan, Jr.d, and Roger Y. Tsiena,c,e,1
aDepartment of Pharmacology,bNeurosciences Graduate Group,dDivision of Biological Sciences,eDepartment of Chemistry and Biochemistry, andcHoward
Hughes Medical Institute, University of California at San Diego, La Jolla, CA 92093
Contributed by Roger Y. Tsien, December 21, 2011 (sent for review November 26, 2011)
Fluorescence imaging is an attractive method for monitoring
neuronal activity. A key challenge for optically monitoring voltage
is development of sensors that can give large and fast responses
to changes in transmembrane potential. We now present fluores-
cent sensors that detect voltage changes in neurons by modula-
tion of photo-induced electron transfer (PeT) from an electron
donor through a synthetic molecular wire to a fluorophore. These
dyes give bigger responses to voltage than electrochromic dyes,
yet have much faster kinetics and much less added capacitance
than existing sensors based on hydrophobic anions or voltage-
sensitive ion channels. These features enable single-trial detection
of synaptic and action potentials in cultured hippocampal neurons
and intact leech ganglia. Voltage-dependent PeT should be
amenable to much further optimization, but the existing probes
are already valuable indicators of neuronal activity.
thus complements traditional electrophysiological measurements
(1, 2). Ca2+imaging is the most popular of such techniques,
because the indicators are well-developed (3–6), highly sensitive
(5, 6), and genetically encodable (7–13), enabling investigation of
the spatial distribution of Ca2+dynamics in structures as small as
dendritic spines and as large as functional circuits. However,
because neurons translate depolarizations into Ca2+signals via
a complex series of pumps, channels, and buffers, fluorescence
imaging of Ca2+transients cannot provide a complete picture of
electrical activity in neurons. Observed Ca2+spikes are tempo-
rally low-pass filtered from the initial depolarization and provide
limited information regarding hyperpolarizations and subthreshold
events. Direct measurement of transmembrane potential with
fluorescent indicators would provide a more accurate account
of the timing and location of neuronal activity. Despite the prom-
ise of fluorescent voltage-sensitive dyes (VSDs), previous classes of
VSDs have each been hampered by some combination of insen-
sitivity, slow kinetics (14–16), heavy capacitative loading (17–21),
lack of genetic targetability, or phototoxicity. Two of the more
widely used classes of VSDs, electrochromic and FRET dyes,
illustrate the problems associated with developing fast and sensi-
tive fluorescent VSDs.
Electrochromic dyes respond to voltage through a direct in-
teraction between the chromophore and the electric field
(Scheme 1A). This Stark effect leads to small wavelength shifts in
the absorption and emission spectrum. Because the electric field
directly modulates the energy levels of the chromophore, the
kinetics of voltage sensing occur on a timescale commensurate
with absorption and emission, resulting in ultrafast (fs to ps)
hypso- or bathochromic shifts many orders-of-magnitude faster
than required to resolve fast spiking events and action potentials
in neurons. This small wavelength shift dictates that the fluo-
rescence signal can be best recorded at the edges of the spec-
trum, where intensity varies most steeply as a function of
wavelength. The largest linear responses are −28% ΔF/F per 100
mV (22), although more typical values are ∼10% per 100 mV
(23, 24). Photo-induced electron transfer (PeT)-based Ca2+
probes, such as fluo-3, give ΔF/F values of up to 150% for action
potentials in cultured hippocampal neurons (25). Therefore,
luorescence imaging can map the electrical activity and
communication of multiple spatially resolved neurons and
although electrochromic dyes can keep pace with fast voltage
oscillations in neurons, their insensitivity limits the systems in
which these dyes can successfully report on voltage changes.
FRET-based voltage sensors use lipophilic anions that in-
tercalate into the cellular membrane and distribute between the
inner and outer leaflets depending upon the transmembrane
potential (Scheme 1B). The Nernstian distribution is monitored
by a second fluorophore immobilized on one side of the mem-
brane, which undergoes FRET preferentially with the mobile
anions on the same side of the membrane. Translocation of the
lipophilic anion through the lipid bilayer governs the kinetics of
voltage sensing, which can be in the millisecond range. Although
these two-component systems can give large changes in intensity
(5–34%) (21) or ratio (80% per 100 mV) (15), the slow trans-
location of mobile charges in the plasma membrane introduces a
capacitative load and hampers the ability of the reporter to
monitor fast changes.
To combine the best features of electrochromic and FRET-
based VSDs, we have now tested a unique mechanism for voltage
sensing, PeT through molecular wires. In these PeT sensors,
a fluorescent reporter connects to an electron-rich quencher via
a molecular wire, which minimizes the exponential distance de-
pendence of intramolecular electron transfer (26) and allows
efficient electron transfer over a major fraction of the thickness
of the plasma membrane. At resting or hyperpolarized poten-
tials, the transmembrane electric field promotes electron transfer
from the quencher to the excited-state fluorophore through the
molecular wire, quenching fluorescence (Scheme 1C). De-
polarization reverses the electric field, hinders electron transfer,
and brightens fluorescence (27), just as Ca2+binding dequenches
indicators like fluo-3 (28). Electron transfer occurs within pico-
to nanoseconds after photon absorption and returns to its initial
state within a microsecond (26, 29), slower than the electro-
chromic mechanism but essentially instantaneous on a biological
timescale. Because electron transfer reverses quickly and is
driven by photon absorption rather than membrane potential
changes, capacitative loading should be negligible, as calculated
in the SI Appendix. A full electronic charge traverses a Marcus-
type thermal activation barrier to sense a large fraction of the
membrane voltage, making voltage sensitivity high (30). Quench-
ing of the fluorescent reporter by the electron-rich donor modu-
lates the fluorescence quantum yield independent of wavelength,
permitting efficient use of photons for excitation and emission,
allowing lower light levels or dye concentrations to be used. We
report here the design, synthesis, and application of the Volta-
geFluor (VF) family of fluorescent sensors as molecular wire
PeT-based probes for voltage imaging in neurons.
Author contributions: E.W.M., W.B.K., and R.Y.T. designed research; E.W.M., J.Y.L., E.P.F.,
and P.A.S. performed research; E.W.M., E.P.F., W.B.K., and R.Y.T. analyzed data; and
E.W.M. and R.Y.T. wrote the paper.
The authors declare no conflict of interest.
Freely available online through the PNAS open access option.
1To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| February 7, 2012
| vol. 109
| no. 6 www.pnas.org/cgi/doi/10.1073/pnas.1120694109
Design and Synthesis of VF Sensors. Our initial voltage sensors in-
corporate dichlorosulfofluorescein as a membrane-impermeant
fluorophore, a p-phenylenevinylene (PPV) molecular wire, and N,
N-dimethyl- or dibutylaniline as an electron-rich quencher
(Scheme 2). VF1.4.Cl comprises 2,7-dichlorosulfofluorescein con-
nected via one vinylene unit to dibutylaniline (hence VF1.4.Cl).
VF2.4.Cl adds a second PPV unit, and VF2.1.Cl features the same
configuration, with methyl substituted in place of butyl groups.
Correct positioning of the fluorophore-wire donor within the
membrane is vital to take advantage of the vectoral nature of the
transmembrane electric field and electron transfer. First, the lon-
gitudinal axis of the molecular wire must be normal to the plane
of the plasma membrane, to sample the full electric field. Second,
dye molecules must all align in the same direction to avoid can-
at the extracellular leaflet of the membrane ensures fluorescence
brightening upon depolarization; the opposite orientation of PeT
would give fluorescence quenching upon depolarization.
The negatively charged sulfofluorescein will preclude dye in-
ternalization and force an orientation in which the fluorophore
adsorbs to the outer leaflet of the plasma membrane, with the
lipophilic molecular wire and alkyl aniline dangling into the lipid
bilayer. As an intervening spacer, PPV molecular wires are an
tractability, and demonstrated ability to conduct current through
lipid bilayers (31). Anilines are common PeT donors and the di-
alkyl groups should enhance uptake into the plasma membrane.
A modular synthetic design both allows for rapid generation of
the voltage sensors and enables future derivatization (Scheme 2
and SI Appendix). Coupling of the molecular wire styrene unit 1,
available in one step from 4-di-butylaminobenzaldehyde, with
iodo-functionalized dichlorosulfofluorescein 2 via a Pd-catalyzed
Heck reaction gives VF1.4.Cl in good yield. An analogous re-
action with molecular wire 3, available in two steps from 1, gives
VF2.4.Cl in 70% yield. A parallel reaction beginning from styrene
4 furnishes VF2.1.Cl in good yield. All dyes feature emission and
excitation profiles typical of dichlorofluoresceins (VF1.4.Cl: λmax=
521 nm, ε = 93,000 M−1·cm−1, λem= 534 nm, Φ = 0.24; VF2.4.Cl:
directly with the electric field. Absorption of a photon significantly alters the excited state molecular dipole, which at hyperpolarizing potentials is stabilized
(Left). At depolarizing potentials the charge shift inverted state is destabilized (Right). Changes in the energy levels of the chromophore result in small
spectral shifts in the emission of the dye. (B) FRET-pair voltage sensors use lipophilic anions (red), which partition in a voltage-dependent fashion on the inner
or outer leaflet of the membrane. Depolarization causes translocation of the anion, which can now quench the fluorescence of an immobilized fluorophore
(green). (C) Molecular wire PeT VSDs depend upon the voltage-sensitive electron transfer from an electron-rich donor (orange) through a membrane-
spanning molecular wire (black) to a fluorescent reporter (green). At hyperpolarizing potentials, the electric field is aligned antiparallel to the direction of
electron transfer, resulting in efficient PeT and quenched fluorescence (Left). Depolarization aligns the electric field in the direction of PeT, decreasing the
rate of electron transfer and increasing fluorescence (Right).
Mechanisms of fluorescent voltage sensing. (A) Electrochromic VSDs sense voltage through the Stark effect, whereby the chromophore interacts
Scheme 2. Synthesis of VF probes.
Miller et al.PNAS
| February 7, 2012
| vol. 109
| no. 6
λmax= 522 nm, ε = 97,000 M−1·cm−1, λem= 536 nm, Φ = 0.054;
0.057, 5 mM sodium phosphate, pH 9, 0.1% Triton X-100) (SI
Appendix, Fig. S1). The dibutyl (VFx.4.Cl) dyes stain the cell
membranes of HEK293 cells when loaded at a concentration of
2 μM for 15 min at 37 °C in buffer with 0.1% DMSO as cosolvent
(Fig. 1A, and SI Appendix, Fig. S2). VF2.1.Cl requires even lower
dye concentrations (100 nM) and gives bright staining of HEK cell
membranes, which is likely to be because of the greater aqueous
solubility of VF2.1.Cl compared with VF2.4.Cl in aqueous solution
(SI Appendix, Fig. S2). The membrane retention of the second
generation dyes (VF2.x.Cl) isin contrast to di-4-ANEPPS, which at
the same loading conditions, shows significant uptake into internal
membranes. The bleach rates of the probes were tested in HEK
Cl, VF2.4.Cl, and VF2.1.Cl at 7 W/cm2were measured to be 3.9 ±
0.1 × 10−2s−1, 1.8 ± 0.1 × 10−2s−1, and 8.0 ± 0.1 × 10−3s−1, re-
spectively. These results are two-, four-, and ninefold smaller than
conditions, of 6.9 ± 0.1 × 10−2s−1.
Characterization of the Voltage Response of VF Sensors. We char-
acterized the voltage sensitivity of all three indicators by making
tight-seal whole-cell recordings of HEK293 cells stained with the
VF sensors. Cells were voltage-clamped at −60 mV holding po-
tential and sequentially stepped to depolarizing and hyper-
polarizing potentials at 20-mV increments (Fig. 1B). For all three
dyes, depolarizing steps produced fluorescenceincreases, whereas
hyperpolarizing steps produced fluorescence decreases, in keep-
ingwiththeproposedPeTmechanism. Thefluorescence response
is linear over the range of −100 mV to +100 mV (Fig. 1C), with
25 ± 1% for VF2.4.Cl. This statistically significant increase in
voltage sensitivity (P < 0.05, two-tailed Student t test) is expected
upon increasing the length of the molecular wire, and is 2.5- to
4-times more sensitive than di-4-ANEPPS, which, in our hands,
gives sensitivities of between 6% and 10% ΔF/F per 100 mV.
VF2.1.Cl shows fluorescence increases upon depolarization sim-
ilar to VF2.4.Cl, with a voltage sensitivity of 27 ± 1% per 100 mV.
This value is not significantly different from the sensitivity of
VF2.4.Cl, suggesting that voltage sensitivity is largely determined
by the length of the molecular wire and that the small increase in
electron donating ability of the butyl relative to the methyl groups
makes only a relatively small contribution to the increased voltage
sensitivity of VF2.1.Cl. To investigate the speedof response of VF
probes, we again made whole-cell recordings of cells stained with
VF2.4.Cl, applied 100-mV depolarizing steps from a holding po-
tential of −60 mV, and recorded both the electrophysiological
signals and the optical signals. Fitting the electrophysiological and
and end of the pulse (τON, phys= 139 ± 0.2 μs, τON, optical= 138 ±
14μs,τOFF, phys=142±0.4μs,τOFF, optical=147±19μs),showing
that VF2.4.Cl and related sensors do not introduce any detectable
lag in their fluorescence response to voltage, consistent with
a PeT-based mechanism for voltage sensing (Fig. 2 A and B).
The fluorescence response of VF2.4.Cl to voltage changes is
insensitive to the excitation wavelength, as is true for PeT-based
probes, such as fluo-3 and Calcium Green-1. We assayed the
wavelength dependence of VF2.4.Cl by changing the excitation
wavelength in 5-nm steps and determined that the fluorescence
response of VF2.4.Cl to a 100-mV depolarization from a holding
potential of −60 mV varied only about 15% when testing
wavelengths from 445 to 500 nm (Fig. 2C). In comparison, di-4-
ANEPPS varies by nearly 100% over its excitation spectrum (32),
and the PeT-based Ca2+sensor, Calcium Green-1, varies by
∼20% (SI Appendix, Fig. S3). These comparisons show that the
wavelength independence of the voltage sensitivity is more
consistent with PeT than a wavelength-shifting mechanism, such
as electrochromism or solvatochromism (a wavelength shift be-
cause of alteration in local solvation).
PeT-based molecular wire sensors do not affect neuronal ex-
citability by capacitative loading. We injected hyperpolarizing
current into the Retzius cells of leech ganglia preparations and
compared the time constants for these voltage steps in ganglia
under different dye loading conditions. The ganglia were stained
with dye at three times the working concentration [either VF2.1.
Cl or diSBA-C4-(3) (14)] and compared with unloaded cells. The
presence of the translocating dye oxonol 413 substantially
increases the capacitative load on the membrane, as measured by
the increase in the RC time constant for the hyperpolarizing
injection (Fig. 2D). On the other hand, ganglia loaded with
VF2.1.Cl show no difference from control cells, demonstrating
that molecular wire sensors place negligible capacitative load on
the cell (Fig. 2D), confirming the predictions of the SI Appendix.
Detection of Action Potentials by VF2.4.Cl in Mammalian Neurons. To
assess whether VF probes can detect action potentials in single
trials, we used cultured rat hippocampal neurons. Bath applica-
tion of 2 μM VF2.4.Cl showed bright cell staining limited to the
cell membranes of neurons and their support cells (Fig. 3A). We
then injected current into a neuron under whole-cell patch-clamp
mode to trigger single action potentials and used a high-speed,
back-illuminated EMCCD camera to track fast optical signals
from VF2.4.Cl, enabling us to resolve action potentials in neurons
in single sweeps (Fig. 3B). The optical trace matched the physi-
ology trace and gave about a 20% ΔF/F increase in fluorescence
and a 16:1 signal-to-noise ratio (SNR) in a single trial. The fact
that VF2.4.Cl detected action potentials without spike-timed av-
eraging suggests the possibility of measuring spontaneous action
potentials in neurons at sites away from the recording pipette.
Monitoring Spontaneous Activity in Leech Ganglia with VF2.1.Cl. A
more stringent test of the usefulness of the PeT-based VSD is to
determine whether it can accurately measure subthreshold ac-
tivity in heterogeneous preparations. For this test we used leech
ganglia, because their neurons have been well studied using both
VF2.4.Cl fluorescence during a series of voltage steps to +100 or −100 from a holding potential of −60 mV (40-mV increments). (C) Fractional changes in VF2.4.
Cl fluorescence from B plotted against membrane potential for voltage changes from a holding potential of −60 mV. Each datapoint represents three to four
separate measurements. Error bars are SEM.
Characterization of VF sensors in HEK cells. (A) Confocal image of HEK 293 cells stained with 2 μM VF2.4.Cl. (Scale bar, 20 μm.) (B) Fractional changes in
| www.pnas.org/cgi/doi/10.1073/pnas.1120694109Miller et al.
electrophysiological and other VSD recordings (14, 33–35). We
isolated a midbody ganglion and stained it with VF2.1.Cl for 15
min at 22 °C (Fig. 3C). Insertion of a sharp electrode (25 MΩ)
into a Retzius cell to enabled recording of its spontaneous ac-
tivity while simultaneously recording the fluorescence signals
from the same cell. When sampling at a rate of 50 Hz, the optical
recording (Fig. 3D, red trace) faithfully followed the sub-
threshold fluctuations in the electrical recording (black trace).
The optically recorded spikes are truncated as a result of
undersampling the optical signal; sampling at a higher rate (722
Hz) fully resolved the action potentials, but introduced a signifi-
cant amount of sampling noise (SI Appendix, Fig. S4). Although
the action potential was subsampled at just 50 Hz, there is still a
reliable transient in the optical trace that indicates the time when
an action potential occurs, which is often what is needed.
The PeT-based VSDs show significant improvement in speed
and accuracy compared with FRET-based VSDs previously used
for leech recordings (33, 36–38), which in turn had superseded
electrochromic dyes (14). The improvement in the recording of
membrane potential fluctuations is not the result of a greater
sensitivity (the ΔF/F for both the FRET and PeT dyes is about
10% per 100 mV in leech recordings), but to a greater SNR. The
PeT-based VSD produces a much brighter signal, one that is well
above the photon noise levels of the dye and the dark noise level
of the camera. Increasing the concentration of the FRET-based
dye does increase its SNR, but the consequent increase in the
cell’s capacitance (Fig. 2D) makes the dye useless for recording
either action potentials or synaptic potentials. Tests of the tox-
icity and bleaching of the PeT-based VSD similar to those per-
formed on the FRET-based dyes (14) show that the PeT-based
VSD has a slower rate of bleaching and is less toxic than the
FRET-based dyes. Hence, considering all measures, the PeT-
based VSD performs better than the FRET-based dyes (39).
Optimal VSDs would have large, fast responses to changes in
voltage, place little or no capacitative load on the membrane,
photobleach slowly with minimal photodynamic damage, and
would be synthetically tractable for rational chemical modifica-
tion and genetic targetability. We believe that the VF family of
PeT-based probes surpass previous VSD classes by these criteria.
The three PPV molecular wire, PeT-based molecules we tested
(VF1.4.Cl, VF2.4.Cl, and VF2.1.Cl) exhibit good membrane
staining and 20–27% ΔF/F per 100-mV increases in fluorescence
upon depolarization in HEK cells. These molecules possess the
fast kinetics (τON/OFF<<140 μs) and wavelength-independent
voltage sensitivity consistent with a PeT mechanism for sensing
voltage. Measurements of capacitance in leech neurons show
that an insignificant amount of capacitative load is placed on the
membrane. The advantages of PeT-based dyes over both elec-
trochromic and FRET-based methods for optical voltage sensing
are described below and summarized in Table 1. A fourth
technique, making use of genetically encoded voltage sensors,
offers a promising method for optically monitoring voltage
changes because the fluorescent proteins can be targeted to cells
of interest, thereby increasing the SNR of the fluorescence re-
sponse. In practice, however, fluorescent protein voltage sensors
suffer from low sensitivity [0.5% (40) to 10% Δ F/F per 100 mV
(41)], nonlinear responses (42) and slow kinetics (tens to hun-
dreds of milliseconds). Newer efforts have made use of proton
translocation within bacterial rhodopsins (43), but although
these show large voltage sensitivities, the response time is still in
the millisecond range, quantum efficiencies are very low, and
their expression limited to prokaryotic systems. Voltage-driven
translocation of ions through the membrane will generally add
much more capacitative load than electron translocation during
transient excited states (44).
VSDs using a PeT-based molecular wire approach should be
highly sensitive. Because a full electronic charge travels through
a substantial fraction of the transmembrane voltage (11 Å for
VF1.4.Cl, 17 Å for VF2.4.Cl and VF2.1.Cl, or 37% and 57% of
the 30 Å low-dielectric constant core of the plasma membrane)
the change in driving force for PeT is large. For example, a 100-
mV depolarization changes the PeT driving force by 0.05 eV
(one electron × half of 100-mV potential, or 0.05 V). Because
PeT is a thermally controlled process, the value of 0.05 eV is
large relative to the value of kT at 300 K (0.026 eV), yielding
a large dynamic range between the rates of PeT at resting and
depolarized potentials. FRET-based VSDs will have similar
sensitivities; lipid-soluble mobile anions transverse distances
calculated to be between 0.4 (16) and 0.6 (45) of the total
membrane width, resulting in ΔG of ∼0.05 eV for 100-mV de-
polarization, compared with a kT of 0.026 eV for the thermally
In contrast, electrochromic dyes have smaller ΔG values, 0.003
(46) to 0.02 (47) eV, and larger comparison energies. Because
the interaction is a photochemically controlled process, the en-
ergy of the exciting photon is the comparison energy, which is
1.5–2 eV for dyes in the blue-to-green region of the spectrum.
Therefore, PeT and FRET dyes have large changes in energy
versus their comparison energy (0.05 eV vs. 0.026 eV), giving
high sensitivities; electrochromic dyes have small changes com-
pared with the excitation photon (0.003–0.02 eV vs. 2 eV),
producing low voltage sensitivity.
The nature of the PeT mechanism also predicts that the ki-
netics of voltage sensing will be fast; forward electron transfer
occurs on the nanosecond timescale as fluorescence is quenched,
and back-electron transfer completes the cycle and occurs on a
microsecond timescale or faster, meaning that the slow step,
electron-hole recombination, finishes a full three orders-of-
magnitude faster than an action potential. Electrochromic dyes
tance of the VF2 fluorescence response. (A) Rising edge of a 100-mV depo-
larizing step from −60 mV in HEK cells stained with VF2.4.Cl. (B) Falling edge
of the same step. Black, solid trace is the integrated current measured
electrophysiologically; red points are the optical recording. Time constants
are calculated by fitting a monoexponential equation to each side of the
step. Traces are the average of 100 sequential trials. (C) Voltage sensitivity
vs. excitation wavelength. The normalized response of VF2.4.Cl to a 100 mV
depolarization from −60 mV in HEK cells is plotted in red, and the excitation
spectrum in HEK cells is the dotted black line. Error bars are SEM for n = 3
experiments. (D) Measurement of capacitative loading in leech Retzius cells.
Traces show the normalized voltage decay following hyperpolarizing cur-
rent injection into Retzius cell stained with 3× VF2.1.Cl (red trace), 3× oxonol
413 (black trace), or nothing (gray trace). (Inset) An expanded time scale
revealing no difference between cells stained with VF2.1.Cl and control cells.
Characterization of the speed, wavelength sensitivity, and capaci-
Miller et al.PNAS
| February 7, 2012
| vol. 109
| no. 6
display even faster kinetics, as forward charge shift occurs with
absorbance, on the femtosecond time scale, and resolves itself
upon emission of a photon, enabling these dyes to keep time with
the fastest spiking neurons. FRET pair VSDs depend upon the
migration of a lipophilic anion through an unstirred lipid bilayer
and display kinetics in the millisecond-to-second time regime,
limiting their application to monitoring only slow transients.
Because PeT shuttles an electron across the membrane and
back on a microsecond or faster timescale, driven by photons
rather than membrane potential changes, no capacitative loading
should be observed. The same holds true for electrochromic dyes,
which transfer electrons on even faster time scales. One disad-
vantage of electrochromic dyes is that they require the entire
voltage-sensing chromophores to be rigid to enable π orbital
overlap, quantum yield, efficient charge transfer, and maximiza-
tion of voltage sensitivity (22). Such rigidity hinders synthesis and
water solubility and may explain why electrochromic dyes are not
improved by lengthening their chromophores. PeT probes do not
require the entire molecular wire to be rigidly coplanar, and
synthesis of longer wires is feasible.
PeT-sensing allows the entire emission spectrum to be used for
monitoring voltage, because the quenching mechanism alters the
ΦFl, decreasing the brightness of the dye, and does not shift the
wavelength as do electrochromic methods. Because photons are
not wasted, this allows lower intensity light to be used in
experiments, reducing phototoxicity and increasing the duration
of experimental procedures. The performance of electrochromic
dyes has plateaued over four decades of development. Excitation
at the far-red edge of the spectrum gives voltage sensitivities
ranging from −35% to −52% ΔF/F per 100 mV; however, at the
edge of the spectrum, the intensity is far below the peak and the
voltage response becomes nonlinear (48).
Several limitations of the VF dyes remain to be addressed. VF
derivatives are not yet genetically targetable. The sensors are
readily taken up by the cell membranes of all tissue, increasing
nonresponsive background fluorescence and decreasing the SNR.
For heterogeneous preparations, such as intact leech ganglia and
brain slices, thisbecomes an increasingly important issue, and one
method to address this concern is through the genetic targeting of
VSDs. VF sensors lend themselves to chemical derivatization,
and efforts are underway to modify VF probes for targeting to
genetically defined circuits of neurons.
voltage as hoped. Our first derivatives show sensitivities ranging
from 20–27% ΔF/F per 100 mV, and the most sensitive of existing
electrochromic dyes exhibit −28% ΔF/F per 100-mV sensitivities
in the linear range (47). Although it is encouraging that the first
derivatives display sensitivities on a level approaching the most
Table 1. Summary of VSD attributes
Nature of translocating charge
Forward charge shift occurs when
Reverse charge shift occurs when
Photon emission or
Lipid soluble anion
∼0.5Fractional charge x Fraction
of total voltage
Δ energy for 100 mV ΔV
Extended rigid fluorophore needed?
Use full ex/em band
Sensitivity Δ F/F per mV
Rat hippocampal neurons stained with 2 μM VF2.4.Cl for 15
min show strong membrane staining. (Scale bar, 20 μm.) (B)
VF2.4.Cl can detected evoked action potentials in rat hip-
pocampal neurons in single trials. The black trace is the
recorded electrophysiology signal. Individual points repre-
sent the optical signal from VF2.4.Cl captured with a high
speed EMCCD camera at a rate of 2 kHz. (C) Optical im-
aging of spontaneous activity in leech Retzius cells using
the dye VF2.1.Cl. Desheathed midbody leech ganglion
stained with 200 nM VF2.1.Cl for 15 min. Pixels within the
region of interest (red circle around a single Retzius cell
body) were averaged in each frame to produce the optical
trace. (Scale bar, 25 μm.) (D) Simultaneous optical and
electrophysiological recording of spontaneous activity in
cell from C. The red trace is the hi-pass filtered VF2.1.Cl
signal, sampled at 50 Hz. The black trace is the electro-
physiological recording, sampled at 10 kHz. The optical
trace shows near-perfect matching of the subthreshold
membrane potential and a clear detectable signal in-
dicating action potentials. Action potentials have variable
amplitudes in the optical traces because of the relatively
slow optical sampling rate (SI Appendix, Fig. S4).
VF2 dyes resolve action potentials in neurons. (A)
| www.pnas.org/cgi/doi/10.1073/pnas.1120694109Miller et al.
sensitive electrochromic dyes, we believe ample chemical space
exists for improving the voltage sensitivity of molecular wire
efficiency of PeT can be rationally tuned (49) by altering the
electron affinities of the donor, wire, and acceptor to maximize
the fluorescence turn-on in response to depolarizations. Addi-
tionally, extending the molecular wire to span an even greater
distance across the plasma membrane should increase sensitivity
as the transferred electron samples more of the electric field. The
modular nature of the VF synthesis allows for rapid interchange
of coupling partners to quickly assemble and assess the voltage
sensitivity of an array of compounds.
In summary, we present a unique method for monitoring volt-
age in neurons based on the voltage-sensitive PeT from an elec-
tron-rich donor to fluorescent reporter attached via a membrane-
spanning molecular wire. The VF family of sensors have large,
linear, turn-on fluorescence responses to depolarizing steps (20–
27% ΔF/F per 100 mV), fast kinetics (τ <<140 μs), and negligible
capacitative loading. VF2.4.Cl can detect and resolve evoked ac-
tion potentials in primary culture hippocampal neurons, and
in leech Retzius cells with sensitivity and time-course essentially
identical to the recorded electrophysiology signal. VF sensors
provide a practical alternative to currently available VSDs, and
future derivatives of the molecular wire platform will increase our
ability to optically monitor the temporal and spatial dynamics of
neuronal activity in defined circuits of neurons.
Imaging, electrophysiology, cell culture, leech imaging and electrophysiol-
ogy, and data analysis methods are available in SI Appendix. Theoretical
considerations of capacitative load are included in SI Appendix. Dyes were
synthesized using standard synthetic procedures detailed in SI Appendix.
photometer and Intelligent Imaging Innovations and Photometrics for use of
the Evolve 128 camera. This work was supported in part by the Howard
Hughes Medical Institute and US National Institute of Neurological Disorders
and Stroke Grant R37 NS027177 (to R.Y.T.); National Institutes of Health
Grant MH43396 and National Science Foundation Grant IOB-0523959 (to
W.B.K.); the National Institute of Biomedical Imaging and Bioengineering
for Postdoctoral Fellowship F32 EB012423 (to E.W.M.); and National Institutes
of Health Training Grants EB009380 and MH020002 (to E.P.F.).
1. Scanziani M, Häusser M (2009) Electrophysiology in the age of light. Nature 461:
2. Peterka DS, Takahashi H, Yuste R (2011) Imaging voltage in neurons. Neuron 69:9–21.
3. Poenie M, Alderton J, Tsien RY, Steinhardt RA (1985) Changes of free calcium levels
with stages of the cell division cycle. Nature 315:147–149.
4. Tsien RY, Rink TJ, Poenie M (1985) Measurement of cytosolic free Ca2+ in individual
small cells using fluorescence microscopy with dual excitation wavelengths. Cell Cal-
5. Grynkiewicz G, Poenie M, Tsien RY (1985) A new generation of Ca2+ indicators with
greatly improved fluorescence properties. J Biol Chem 260:3440–3450.
6. Minta A, Kao JP, Tsien RY (1989) Fluorescent indicators for cytosolic calcium based on
rhodamine and fluorescein chromophores. J Biol Chem 264:8171–8178.
7. Nagai T, Sawano A, Park ES, Miyawaki A (2001) Circularly permuted green fluorescent
proteins engineered to sense Ca2+. Proc Natl Acad Sci USA 98:3197–3202.
8. Miyawaki A, et al. (1997) Fluorescent indicators for Ca2+ based on green fluorescent
proteins and calmodulin. Nature 388:882–887.
9. Tian L, et al. (2009) Imaging neural activity in worms, flies and mice with improved
GCaMP calcium indicators. Nat Methods 6:875–881.
10. Mank M, et al. (2008) A genetically encoded calcium indicator for chronic in vivo two-
photon imaging. Nat Methods 5:805–811.
11. Heim N, Griesbeck O (2004) Genetically encoded indicators of cellular calcium dy-
namics based on troponin C and green fluorescent protein. J Biol Chem 279:
12. Palmer AE, et al. (2006) Ca2+ indicators based on computationally redesigned cal-
modulin-peptide pairs. Chem Biol 13:521–530.
13. Horikawa K, et al. (2010) Spontaneous network activity visualized by ultrasensitive Ca
(2+) indicators, yellow Cameleon-Nano. Nat Methods 7:729–732.
14. Cacciatore TW, et al. (1999) Identification of neural circuits by imaging coherent
electrical activity with FRET-based dyes. Neuron 23:449–459.
15. González JE, Tsien RY (1997) Improved indicators of cell membrane potential that use
fluorescence resonance energy transfer. Chem Biol 4:269–277.
16. González JE, Tsien RY (1995) Voltage sensing by fluorescence resonance energy
transfer in single cells. Biophys J 69:1272–1280.
17. Sjulson L, Miesenböck G (2008) Rational optimization and imaging in vivo of a ge-
netically encoded optical voltage reporter. J Neurosci 28:5582–5593.
18. Akemann W, Lundby A, Mutoh H, Knöpfel T (2009) Effect of voltage sensitive fluo-
rescent proteins on neuronal excitability. Biophys J 96:3959–3976.
19. Wang D, Zhang Z, Chanda B, Jackson MB (2010) Improved probes for hybrid voltage
sensor imaging. Biophys J 99:2355–2365.
20. Bradley J, Luo R, Otis TS, DiGregorio DA (2009) Submillisecond optical reporting of
membrane potential in situ using a neuronal tracer dye. J Neurosci 29:9197–9209.
21. Chanda B, et al. (2005) A hybrid approach to measuring electrical activity in geneti-
cally specified neurons. Nat Neurosci 8:1619–1626.
22. Hubener G, Lambacher A, Fromherz P (2003) Anellated hemicyanine dyes with large
symmetrical solvatochromism of absorption and fluorescence. J Phys Chem B 107:
23. Grinvald A (1983) Fluorescence monitoring of electrical responses from small neurons
and their processes. Biophys J 42:195–198.
24. Fluhler E, Burnham VG, Loew LM (1985) Spectra, membrane binding, and potentio-
metric responses of new charge shift probes. Biochemistry 24:5749–5755.
25. Jacobs JM, Meyer T (1997) Control of action potential-induced Ca2+ signaling in the
soma of hippocampal neurons by Ca2+ release from intracellular stores. J Neurosci 17:
26. Davis WB, Svec WA, Ratner MA, Wasielewski MR (1998) Molecular-wire behaviour in
p-phenylenevinylene oligomers. Nature 396:60–63.
27. de Silva AP, et al. (1995) New fluorescent model compounds for the study of pho-
toinduced electron transfer: The influence of a molecular electric field in the excited
state. Angew Chem Int Ed Engl 34:1728–1731.
28. Adams SR (2010) How calcium indicators work. Cold Spring Harbor Protocols 2010:
29. de la Torre G, Giacalone F, Segura JL, Martín N, Guldi DM (2005) Electronic commu-
nication through pi-conjugated wires in covalently linked porphyrin/C60 ensembles.
30. Li LS (2007) Fluorescence probes for membrane potentials based on mesoscopic
electron transfer. Nano Lett 7:2981–2986.
31. Garner LE, et al. (2010) Modification of the optoelectronic properties of membranes
via insertion of amphiphilic phenylenevinylene oligoelectrolytes. J Am Chem Soc 132:
32. Montana V, Farkas DL, Loew LM (1989) Dual-wavelength ratiometric fluorescence
measurements of membrane potential. Biochemistry 28:4536–4539.
33. Briggman KL, Abarbanel HD, Kristan WB, Jr. (2005) Optical imaging of neuronal
populations during decision-making. Science 307:896–901.
34. Salzberg BM, Grinvald A, Cohen LB, Davila HV, Ross WN (1977) Optical recording of
neuronal activity in an invertebrate central nervous system: Simultaneous monitoring
of several neurons. J Neurophysiol 40:1281–1291.
35. Ross WN, Arechiga H, Nicholls JG (1987) Optical recording of calcium and voltage
transients following impulses in cell bodies and processes of identified leech neurons
in culture. J Neurosci 7:3877–3887.
36. Taylor AL, Cottrell GW, Kleinfeld D, Kristan WB, Jr. (2003) Imaging reveals synaptic
targets of a swim-terminating neuron in the leech CNS. J Neurosci 23:11402–11410.
37. Briggman KL, Kristan WB, Jr. (2006) Imaging dedicated and multifunctional neural
circuits generating distinct behaviors. J Neurosci 26:10925–10933.
38. Baca SM, Marin-Burgin A, Wagenaar DA, Kristan WB, Jr. (2008) Widespread inhibition
proportional to excitation controls the gain of a leech behavioral circuit. Neuron 57:
39. Briggman KL, Kristan WB, Jr., Gonzalez JE, Kleinfeld D, Tsien RY (2010) Monitoring
integrated activity of individual neurons using FRET-based voltage-sensitive dyes.
Membrane Potential Imaging in the Nervous System: Methods and Applications, eds
Canepari M, Zecevic D (Springer, New York), pp 61–70.
40. Ataka K, Pieribone VA (2002) A genetically targetable fluorescent probe of channel
gating with rapid kinetics. Biophys J 82:509–516.
41. Perron A, et al. (2009) Second and third generation voltage-sensitive fluorescent
proteins for monitoring membrane potential. Front Mol Neurosci 2:5.
42. Siegel MS, Isacoff EY (1997) A genetically encoded optical probe of membrane volt-
age. Neuron 19:735–741.
43. Kralj JM, Hochbaum DR, Douglass AD, Cohen AE (2011) Electrical spiking in Escher-
ichia coli probed with a fluorescent voltage-indicating protein. Science 333:345–348.
44. Sjulson L, Miesenböck G (2007) Optical recording of action potentials and other dis-
crete physiological events: A perspective from signal detection theory. Physiology
45. Fernández JM, Taylor RE, Bezanilla F (1983) Induced capacitance in the squid giant
axon. Lipophilic ion displacement currents. J Gen Physiol 82:331–346.
46. Loew LM, Bonneville GW, Surow J (1978) Charge shift optical probes of membrane
potential. Theory. Biochemistry 17:4065–4071.
47. Kuhn B, Fromherz P (2003) Anellated hemicyanine dyes in a neuron membrane:
Molecular Stark effect and optical voltage recording. J Phys Chem B 107:7903–7913.
48. Kuhn B, Fromherz P, Denk W (2004) High sensitivity of Stark-shift voltage-sensing
dyes by one- or two-photon excitation near the red spectral edge. Biophys J 87:
49. Ueno T, et al. (2004) Rational principles for modulating fluorescence properties of
fluorescein. J Am Chem Soc 126:14079–14085.
Miller et al.PNAS
| February 7, 2012
| vol. 109
| no. 6